The present invention relates to a communication band control system located in a network, and more particularly to the communication band control system located in an IP network arranged to use an IP (Internet Protocol) as a communication protocol.
Today, it is a matter of course that the advance of the internet and the intranet and the higher performance of hardware, representatively a personal computer, need to run a multimedia application on the internet and the intranet.
As a network medium used for the internet and the intranet, the Ethernet is used, and as an upper protocol an IP (Internet Protocol) is used. The Ethernet and the IP protocol are the band-shared networks. At once, they are the best-effort type networks as well, on which the maximum band is always used for performing the communications without having to depend on the type of the data to be communicated.
On the IP network, therefore, any traffic concerning the WEB applications and the multimedia ones and any mission-critical communication as in an in-company business system may be treated equally so that the traffics and the communications may share the communication band on the same network.
As a result, the file transfer may pressurize the band, so that some disadvantages may take place. For example, the data may be delayed on the mission-critical business application, the communication data as on the WEB application may become an obstacle to securing the communication band, or the fluctuation may take place on the transfer of data such as voice that needs to secure the delay time.
The reason why those disadvantages are brought about is that the IP network uses one shared band as the best effort mode without having to depend on an application and a data type.
In order to overcome those disadvantages, the network segments for data transfer required for guaranteeing the communication quality may be separated from the other network segments in order that a sufficiently wide band may be secured. This remedy needs an additional investment, so that it is not practical.
In light of the foregoing background, the needs on the communication band control technique are rising which technique is for efficiently using the band of the IP network by controlling the communication band to be used for each application and data type.
Today, the following standard techniques on the communication band control has been proposed.
1. RSVP (Resource Reservation Protocol)
RSVP is a signaling protocol that is operated as an upward protocol of TCP/IP and is installed in an application on a terminal and a router. This protocol is executed to issue a message called PATH for reserving a necessary communication band from an application of a requesting terminal to a destination terminal. The PATH message is routed as in the ordinary message and then is sent to the destination terminal. In this routing, all the routers for relaying the PATH message are served to transfer the PATH message to the destination terminal if a requested band is secured and then give back to the requesting terminal a RESV message for representing the band is secured. If the requested band is not secured, a reject message is given back to the requesting terminal. These series of operations make it possible to secure the bands on all the routers on the path leading from the requesting terminal to the destination one.
As noted above, the RSVP is the protocol for reserving the band, which protocol does not cover the implementation of the band. Further, the router located on the communication path is used for managing the band. Hence, the RSVP provides no capability of controlling the band if the requesting terminal and the destination terminal are located on the same segment (subnetwork). In order to overcome this shortcoming, the protocol called SBM (Subnet Bandwidth Manager) that is the expansion of the RSVP is standardized. In the SBM, the band managing server called DSBM (Designated SBM) is contained in the subnet. The PATH message that is sent out of an application on the terminal is transferred to the DSBM if the destination is located in the subnet. The DSBM is served to determine if the band requested by the received PATH message can be secured and then give back the RESV message as a response. It means that the band management within the subnet is executed by the DSBM.
2. DiffServ (Differentiated Services)
The DiffServ is a priority controllable protocol. The application on the terminal is executed to bury a priority to an IP protocol header (DS field) in correspondence with the priority classes categorized according to the significance of the data. Then, a relaying device such as a router is served to recognize the priority and then transfer a packet with a higher priority in order.
The packet with a lower priority is buffered in the relaying device such as a router, so that the packet is delayed or discarded. In place, the flow control may be executed such that the transmission from the terminal is slowed down by changing the window size of TCP/IP so that the buffer may not overflow with a great deal of packets.
In addition, for setting a priority, the priority control on the data link level (Ethernet) is made possible by IEEE802.1p.
For efficiently transmitting the data in the network, the data is transmitted with a large band traffic if the network has enough capacity, while the data is transmitted with a small band traffic if the network does not have enough capacity. That is, it is preferable that the band control may be executed according to the band traffic of the network to be secured when the data is to be transmitted.
The DiffServ (that uses the DS field of the IP protocol or the priority of IEEE802.1p) is executed to determine the priority of the data in advance and then transfer the packet with a higher priority in order. It means that the DiffServ has difficulty in determining the band traffic (transmission data amount per a unit time) on the usage of the band for each communication flow.
On the other hand, the RSVP is designed in view of reserving the band. It can implement the process of securing the band from the sending node to the receiving node but manage the band information for each flow (which band information means the information required for controlling the band) concentratively in the router or the DSBM. It means that the DSBM server is required to be located in the subnet if the RSVP is applied to the band control (traffic control) within the subnet.
In order to make use of the communication capacity in the subnetwork as efficiently as possible, it is necessary to grasp what kind of data flow takes place in the subnetwork (band traffic of the network) and separate the data whose communication quality is to be guaranteed from the data whose communication quality is not to be guaranteed and that is communicated through an idle band when managing and controlling the data to be communicated. If a relaying device such as a router is located in the network, by monitoring the packets passing through the router, it is possible to grasp the band information for each flow and easily grasp the current band traffic. However, the technique is unknown which may apply to the band control in the subnet having no router where all the data packets are passed.
Further, the RSVP needs to make all the terminals be at the same level within the subnet and monitor (capture) all the packets on a leased terminal for the purpose of obtaining the band information from the packets flowing through the network. If the subnet such as 10Base5 or a repeater hub is implemented by one segment, it is possible to obtain the band information by monitoring all the packets, while in the network where the subnet is divided into plural segments by a switching hub, the collection of all the packets within the subnet cannot be implemented by one terminal by means of the particular function of the switching hub, that is, the function that the communication closed inside the segment is not transferred to another segment.
In order to highly efficiently control a band traffic of a communication whose quality is to be guaranteed, as stated above, it is becoming necessary to grasp the band traffic and suppress to a minimum a difference between the presumed band traffic and the actual band traffic flowing through the network when the band control is executed.
The packets flowing through the network are classified into communication data for an application (effective communication data) and various protocols control data including the data retransmitted by a recovery protocol. In particular, it is known that though the latter data is inevitably brought about as an overhead of the former data, the data amount greatly depends on the communication quality. This is obvious because the degraded network communication quality causes data loss and thus the data is to be retransmitted for recovering the data loss.
In order to make use of the restricted band of the network as efficiently as possible, therefore, it is essential to control the communication quality so that in particular the data retransmitted by the recovery protocol does not press the band of the effective communication data. That is, it means that it is essential to control the communication quality so that the actual band traffic of the network is made equal to the band traffic of the effective communication band.
In particular, under the circumstances of a wireless LAN or mobile system, the communication quality may be degraded depending on the place of use. Hence, the communication quality is required to be corrected at each place of use.